Hung-yi Lee's 2022 ML course: all the homework, none of the tuition

What it does The repository houses fifteen homework assignments from National Taiwan University’s 2022 spring machine learning course led by Hung-yi Lee. Each unit bundles a Jupyter notebook and PDF slides targeting one specific technique—regression, CNNs, transformers, GANs, reinforcement learning, and others—offering a semester-long tour through applied deep learning.

The interesting bit Instead of a monolithic framework, the repo treats each homework as a self-contained snapshot of a single concept. It functions as a curated index of what mattered in deep learning circa 2022, assembled by the course staff rather than scraped together by outsiders.

Key highlights

• Fifteen distinct topics, from basic regression to meta-learning and network compression.
• Every assignment ships with both executable *.ipynb notebooks and companion PDF slides.
• Official course materials, not a third-party rewrite.
• README links directly to lecture videos on the instructor’s YouTube channel (where available).
• Topic list tracks the modern deep learning syllabus: Self-Attention, Transformers, Explainable AI, Adversarial Attack, Life-Long Learning.

Caveats

• The README is strictly a table of contents; it offers no dependency lists, hardware notes, or dataset pointers for any assignment.
• Most homework entries beyond HW4 omit direct video links in the README.

Verdict Bookmark this if you are self-studying deep learning and want a structured, topic-by-topic progression through the 2022 curriculum. Look elsewhere if you need a unified library or production-ready pipelines.